In response to the points raised by Weidinger and colleagues:
Our study was not an equivalence/noninferiority trial. The primary objective and statistical analysis were prespecified in the protocol and are reported in the article as stated. Equivalence methods were not used in these analyses. Equivalence was mentioned solely in the context of the sample size calculation, which assessed the length of the 2-sided 95% confidence interval (CI) for the risk differences under various possible incidence rates for test and reference. No formal noninferiority or equivalence margins were defined or planned. A post hoc analysis of the CI for the adverse events (AEs) in Table S3 showed that there are no differences in the incidence of these AEs between treatment groups.
The PETITE study was open-label and had a “real world” design, which allowed the use of low- to medium-potency TCS according to their label in the respective country. The TCS that were used in each of the 28 countries involved in our study depended on which TCS were locally approved for use in infants.
Additional subgroup analyses of the PETITE study data are being carried out and will be reported separately.
No statistical testing of efficacy was planned. The analyses were based on observed values. Patients with a missing evaluation at a visit were not included in the efficacy data summaries. The results were presented for all intent-to-treat patients.
Skin atrophy was not measured and probably not of major importance in our study, because there was only intermittent use of low- and medium-potency TCS.
The analyses of relative risk for serious AEs and serious infections presented by Weidinger et al were not planned in our study. The crude incidences of these events were presented in our article as standard safety-reporting measures. The primary safety analyses were conducted for AEs of primary clinical interest and those with a crude incidence of ≥5% in either treatment group as described in our article. This was not an equivalence trial; therefore, no equivalence margins should be applied to any of the analyses. It is also important to point out that infections were recorded as AEs in our study and confirmation of these events by microbial culture was not mandatory.
Although the incidence density ratio assessed by repeated Poisson regression was higher with pimecrolimus for bronchitis, infected eczema, impetigo, and nasopharyngitis, the crude incidences were only 2% to 4% higher, which we do not consider clinically significant or a convincing reason to favor one atopic dermatitis (AD) treatment over another.
A noncorticosteroid alternative such as pimecrolimus is needed for the treatment of AD given concerns that many patients have regarding steroids and their reluctance to use them.1,2 The key point that our data show in this regard is that treatment with pimecrolimus minimizes the need for TCS (to a median of 7 days over 5 years) in a “real world” setting.
We consider the conclusions we have drawn from our study results are appropriate, ie, that both pimecrolimus and TCS are safe for the long-term management of mild-to-moderate AD.
Conflict of Interest:
Dr Sigurgeirsson has been a consultant for Novartis. Dr Luger has collaborated, advised, and lectured for Astellas, Meda, and Novartis.
- Copyright © 2015 by the American Academy of Pediatrics